Title: How to Create Generalizable AI
Speaker: Anima Anandkumar
Date: August 11, 2020
ABSTRACT
Current deep-learning benchmarks focus on generalization on the same distribution as the training data. However, real-world applications require generalization to new, unseen scenarios, domains, and tasks. I'll present key ingredients that I believe are critical towards achieving this, including (1) compositional systems that have modular and interpretable components; (2) unsupervised learning to discover new concepts; (3) feedback mechanisms for robust inference; and (4) causal discovery and inference that capture underlying relationships and invariances. Domain knowledge and structure can help enable learning in these challenging settings. This talk is beginner-friendly and will give a high-level overview of these challenges.
SPEAKER
Anima Anandkumar
Director of ML Research, NVIDIA; Bren Professor, California Institute of Technology
Anima Anandkumar is a Bren Professor at Caltech and Director of ML Research at NVIDIA. She was previously a Principal Scientist at Amazon Web Services. She has received several honors such as the Alfred. P. Sloan Fellowship, NSF Career Award, Young Investigator Awards from DoD, and Faculty Fellowships from Microsoft, Google, Facebook, and Adobe. She is part of the World Economic Forum's Expert Network. She is passionate about designing principled AI algorithms and applying them in interdisciplinary applications. Her research focus is on unsupervised AI, optimization, and tensor methods.
MODERATOR
Michael Zeller
Head, AI Strategy & Solutions at Temasek; ACM SIGKDD Secretary/Treasurer
Michael Zeller is currently Head of AI Strategy & Solutions at Temasek. His passion is to help organizations deepen and accelerate insights from big data through the power of machine learning, predictive analytics, and data science. Hehas more than 20 years of experience as an entrepreneur, executive, and advisor of technology-centric organizations. Prior to joining Temasek, Zeller was the CEO of Dynam.ai, leading its mission for delivering end-to-end artificial intelligence (AI) solutions. Prior to Dynam.AI, he directed strategy and innovation in artificial intelligence for global software firm Software AG. Previously, he was CEO and co-founder of Zementis, a leading provider of software solutions for predictive analytics.
Zeller also serves as Secretary/Treasurer on the Executive Committee of ACM SIGKDD, the premier international organization for data science.
00:00 Intro
5:08 Generalizable AI
16:58 Brain-Inspired Architectures With Recurrent Feedback
26:50 Takeaways
27:46 Neuro-Symbolic Systems for Compositional Reasoning
32:52 Takeaways 2
33:24 AI4Science - Role of Priors
40:59 Takeaways 3
41:38 Unsupervised Learning
46:06 Task Adaptation and Generalization
48:44 Conclusion
